Re: [PATCH 1/5] perf: Add hardware breakpoint address mask

[Jacob Shin <jacob.shin@amd.com>]

On Sun, Apr 21, 2013 at 07:02:02PM +0200, Oleg Nesterov wrote:
> On 04/20, Jacob Shin wrote:
> >
> > On Sat, Apr 20, 2013 at 06:53:34PM +0200, Oleg Nesterov wrote:
> > >
> > > And does attr.bp_len "contribute" to the mask?
> > >
> > > I mean, if bp_len == X86_BREAKPOINT_LEN_8, does this mean that
> > > bp_bp_addr_mask and (bp_bp_addr_mask | 7) have the same effect?
> >
> > Yes it has the same effect.
> 
> OK, thanks...
> 
> So this is the "natural" extension. Given that currently bp_addr
> should be aligned, bp_len could be already bp_mask but I guess it
> is too late to change this, so we need another field.
> 
> Hmm. Perhaps arch_has_hw_breakpoint_addr_mask(void) should be turned
> into arch_validate_hw_breakpoint_addr_mask(bp) which should also
> check that (bp_addr & bp_addr_mask) == 0. But I won't insist.

Yes I can do that .. in that case should the Kconfig
CONFIG_HAVE_HW_BREAKPOINT_ADDR_MASK go away, and in every non-x86
hw_breakpoint.c do:

bool arch_validate_hw_breakpoint_addr_mask(struct perf_event *bp)
{
	return false;
}

?

Or keep CONFIG_HAVE_HW_BREAKPOINT_ADDR_MASK and in
include/linux/hw_breakpoint.h do:

#ifndef CONFIG_HAVE_HW_BREAKPOINT_ADDR_MASK
static inline bool arch_validate_hw_breakpoint_addr_mask(struct perf_event *bp)
{
	return falase;
}
#endif

?

Thanks,